Gas separations may be carried out by a number of methods including distillation at cryogenic temperatures, the use of permselective membranes and by processes that utilize compositions that can reversibly and selectively sorb a component of the gas mixture. For sorption-based separation of air, current commercial technologies utilize zeolite molecular sieves as N.sub.2 -selective adsorbents and carbon molecular sieve (CMS) materials as O.sub.2 -selective adsorbents. These technologies, which are usually employed for the production of enriched nitrogen or oxygen, (rather than very high purity N.sub.2 or O.sub.2) have several inherent limitations which restrict their competitiveness against the cryogenic and membrane separation methods.
Synthetic zeolites reversibly adsorb nitrogen in preference to oxygen. When used for instance in a pressure-swing adsorption (PSA) process for the separation of air, the zeolite bed selectively takes up the nitrogen which is recovered by de-pressurization or evacuation of the bed. The drawback in this separation method is that it is performed inefficiently by adsorbing nitrogen which is the major component of air.
The potential advantages of selective oxygen sorbents have long been recognized and there has been much research effort directed at the synthesis of suitable materials. At the present time carbon molecular sieve (CMS) kinetically oxygen selective adsorbents are used in PSA air separation processes for the production of either enriched N.sub.2 or O.sub.2. Several factors limit the productivity and hence the cost-effectiveness of this technology. Even the most effective current CMS sorbents have a poor working O.sub.2 /N.sub.2 selectivity in the PSA process. The necessarily short cycle times of the PSA process and the limiting oxygen adsorption kinetics lead to a poor utilization of the adsorption bed.
U.S. Pat. No. 4,477,418 discloses solid state transition metal hexacyano compounds (cyanometallates) defined as M.sub.x [M'(CN).sub.6 ].sub.y where M=Sc, Mn, Fe, Co, Ni etc and M' is strictly Cr, Mn, Fe, Co which are selective oxygen sorbents which are taught to be useful in processes for the separation of oxygen. The hexacyanometallate solids can be microporous, containing very small voids within their structures. In certain cases, depending on the specific formula, when the voids are of molecular dimensions the compounds have been described as "molecular sieves" since only molecules that are less than a certain effective diameter are adsorbed within their structures. The experimental data presented in U.S. Pat. No. 4,477,418 show that a number of the listed hexacyanometallates exhibit O.sub.2 versus N.sub.2 adsorption selectivity. Selectivity is seen at short contact times but also, to a lesser extent, at apparent equilibrium conditions. Among the compositions studied there are wide variations in both the time-dependent (i.e. kinetic) and equilibrium values of the oxygen loading, O.sub.2 /N.sub.2 selectivity (ratio of oxygen to nitrogen loading) and in the kinetics of oxygen adsorption. The data show an approximate inverse relationship between the rate of oxygen uptake and the O.sub.2 /N.sub.2 selectivity which is consistent with a molecular sieving or size-selective physical adsorption process, one which is more favorable for entry of the smaller O.sub.2 molecule.
A relatively limited number of solid state chemical O.sub.2 -selective sorbents are known. One of the oldest is the barium oxide/peroxide system disclosed by J. H. Hildebrand, J. Amer. Chem. Soc., 34, 246 (1912), which on the basis of the reversible equilibrium: BaO+1/2O.sub.2 .revreaction.BaO.sub.2 at about 600.degree. C. was once used in an industrial process for the separation of air. U.S. Pat. No. 3,980,763 discloses praseodymium oxide materials which bind O.sub.2, converting it to an oxide (O.sup.2-) ion. The process is temperature/pressure reversible at about 400.degree. C.-500.degree. C., and is said to have the advantage over BaO.sub.2 of not being deactivated by atmospheric carbon dioxide. It is taught in U.S. Pat. No. 4,251,452 that solid manganese phosphine complexes reversibly absorb oxygen, however, the number of reversible oxygen adsorption and desorption cycles that can be obtained appears to be quite limited.
Solid state compositions prepared by an entrapment or encapsulation of a metal complex within the cage of a synthetic zeolite have been shown to function as reversible oxygen sorbents. R. S. Drago, et al, J. Amer. Chem. Soc., 110, 304 (1988) and U.S. Pat. No. 4,830,999 both teach entrapment of the anionic cobalt(II) cyanide (cyanocobaltate(3-)) complexes as ion-paired species: A.sup.+.sub.3 [Co(CN).sub.5 ].sup.3- or possibly A.sup.+.sub.2 [Co(CN).sub.4 ].sup.2- (A.sup.+ is Na.sup.+, Cs.sup.+, etc.) within the pores of a crystalline aluminosilicate zeolite, to yield solid state O.sub.2 -selective sorbents. While the compounds A.sup.+.sub.3 [CO(CN).sub.5 ].sup.3- dissolved in water or polar organic solvents are well known to bind oxygen (giving either superoxo or peroxo complexes, depending on conditions), the O.sub.2 -binding is always considered to be irreversible (Ref. G. A. Kozlov, et al, i Teoreticheskaya Eksperimental'naya Khimiya, 17 (5) 686 (1984)). Thus for example, heating the superoxo complex, [NEt.sub.4 ].sup.+.sub.3 [O.sub.2 Co(CN).sub.5 ].sup.3-, at 120.degree. C. in vacuo gives only a mixture of decomposition products: O.sub.2 , CO.sub.2, butene and other hydrocarbons. The observed reversible binding of O.sub.2 by the same monomeric anionic complex in the zeolite, as described in U.S. Pat. No. 4,830,999, is attributed to as yet uncharacterized interactions between the complex and the walls of the zeolite cavity in which it resides. These interactions significantly change the nature (effectively alter the composition) of the complex such that it becomes reversibly O.sub.2 -binding.
While the entrapment of oxygen-carrier complexes in zeolites affords O.sub.2 -selective solid sorbents, there are significant disadvantages in this technique. Because of the need to incorporate (usually by ion-exchange methods) Co.sup.2+ ions as well as the accompanying organic ligands (e.g. SALEN, CN.sup.-, etc.) in zeolite cages of fixed and usually very small dimensions, and also at the same time retain a certain degree of "openness" within the structure for facile accessibility by O.sub.2, the practical loading level of the active O.sub.2 -binding Co(II) species is often quite small. Thus, as taught by S. Imamura, et al, Langmuir, 1, 326 (1985), in [Co.sup.II (BPY)(TERPY)]-LiY, cobalt complex in LiY zeolite composition, the concentration of Co.sup.II active centers is only 1.05.times.10.sup.-2 mmole/g of zeolite (giving a capacity of about 0.022 cc O.sub.2 /g). In the case of the Co(CN).sub.5.sup.3- /Co(CN).sub.4.sup.2- in zeolite Y sorbent, although a relatively high concentration of Co.sup.+2 (up to 7.1 wt % or 1.2 mmoles/g) can be incorporated, by spectroscopic measurements less than 1% of this cobalt is in an active O.sub.2 -binding configuration (Ref. R. J. Taylor, et al, J. Amer. Chem. Soc., 111, 6610 (1989)). The second drawback of zeolite entrapped metal complex sorbents is their relatively high "background" adsorption capacity for N.sub.2 which limits their O.sub.2 /N.sub.2 selectivity in air separation applications. While the Co(CN).sub.5.sup.3- /Co(CN).sub.4.sup.2- sorbent in zeolite Y at 40 torr pressure has a selectivity (.alpha.O.sub.2 /Ar) of .about.1.3 on the basis of data given in the above reference, the sorbent's oxygen to nitrogen selectivity, (because of the high natural adsorptivity of the latter), is calculated to be less than 1; i.e., about 0.7.
The objective in the art has been to develop easily synthesized solid state metal complex oxygen carriers which have a rapid reactivity and a high reversible equilibrium capacity for oxygen and a relatively low affinity for nitrogen. Additionally, such adsorbents should retain these properties in O.sub.2 recovery applications over a long period of time. Prior to the present invention, no process has been taught which employs adsorbents which meet all of the above qualifications.
S. J. Carter, et al, Inorg. Chem. 25, 2888-2894 (1986) disclose the synthesis of what they believed to have been Li.sub.3 [Co(CN).sub.5 ].multidot.3DMF, although they were unable to purify the material produced in their synthesis reaction. This reference teaches the use of this complex for cyanation reactions, and it is specifically stated that, based upon the research presented in the article, this compound would not be the preferred choice for such reactions. No mention is made of the suitability of this or any similar compound for reversibly binding oxygen. Carter also reported similar findings in a thesis entitled "Synthesis, Characterization and Reactions of New Organocyanocobaltates" Brandeis University, 1988.